Embed Size (px)
Citation preview
1
Potential applications of human viral metagenomics and reference materials: 1
considerations for current and future viruses 2
Tasha M. Santiago-Rodriguez1 and Emily B. Hollister
1 3
1Diversigen, Inc., Houston, TX 77021 4
ABSTRACT 5
Viruses are ubiquitous particles comprised of genetic material that can infect bacteria, archaea, 6
fungi, as well as human and other animal cells. Given that determining virus composition and 7
function in association with states of human health and disease is of increasing interest, we 8
anticipate that the field of viral metagenomics will continue to expand and be applied in a variety 9
of areas ranging from surveillance to discovery, and will rely heavily upon the continued 10
development of reference materials and databases. Information regarding viral composition and 11
function readily translate into biological and clinical applications, including the rapid sequence 12
identification of pathogenic viruses in various sample types. However, viral metagenomic 13
approaches often lack appropriate standards and reference materials to enable cross-study 14
comparisons and assess potential biases which can be introduced at the various stages of 15
collection, storage, processing, and sequence analysis. In addition, implementation of appropriate 16
viral reference materials can aid in the benchmarking of current and development of novel assays 17
for virus identification, discovery, and surveillance. As the field of viral metagenomics expands 18
and standardizes, results will continue to translate into diverse applications. 19
20
Keywords: microbiome, mock communities, reference materials, viral metagenomics, virome 21
Corresponding authors email: [email protected]; [email protected] 22
Address: 2450 Holcombe Blvd. Suite BCMA, Houston, TX, USA 77021 23
AEM Accepted Manuscript Posted Online 11 September 2020Appl. Environ. Microbiol. doi:10.1128/AEM.01794-20Copyright © 2020 Santiago-Rodriguez and Hollister.This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
2
Ongoing potential of viral metagenomics 24
Studies targeting the bacterial fraction of the human microbiome have increased in both 25
number and scope during the last decade. By way of example, a recent PubMed search 26
performed using the keyword “microbiome” demonstrates the exponential growth of microbiome 27
studies since 2000 (Figure 1A). However, the number of studies assessing the viral fraction, or 28
the virome, of various sample types has lagged substantially behind (Figure 1B). Reasons for the 29
dramatically lower number of virome studies include: (i) intrinsic challenges of studying viruses, 30
including their smaller structural and genome sizes relative to bacteria and other 31
microorganisms; (ii) high diversity in genome structure and composition (i.e. dsDNA, ssDNA, 32
dsRNA, and ssRNA); (iii) lack of a universal gene amenable to amplicon sequencing, as in the 33
case of the 16S rRNA gene for bacteria and archaea; and (iv) biases in sequence databases which 34
emphasize viral pathogens or very well-known bacteriophage (i.e., viruses that infect bacteria), 35
many of which have previously been cultured. 36
37
Despite these challenges, viruses are important members of the human microbiome. 38
Viruses are present across the body, at sites including the gut (1), the skin (2), and the oral cavity 39
(3), and they can inhabit body sites and samples types previously thought to be sterile including 40
the bladder (4), blood (5), and cerebrospinal fluid (6). Certain viruses can be acquired through 41
birth and continue to be seeded by the maternal bond (7) and shaped by dietary habits (8), as well 42
as intimate contact (9). Viral communities are associated with disease phenotypes including 43
periodontal disease (10), and Inflammatory Bowel Disease (IBD) (11), and can respond to 44
antibiotic treatment as an indirect response associated with changes in bacterial community 45
composition (12). Viruses, specifically bacteriophages, are now being revisited as a tool to treat 46
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
3
antimicrobial-resistant infections and various disease phenotypes associated with toxin-47
producing bacteria. This was demonstrated in a recent study where bacteriophages infecting 48
cytolysin-producing Enterococcus faecalis strains could ameliorate liver disease severity in 49
patients with alcoholic hepatitis (13). Importantly, a number of phage therapy clinical trials have 50
been conducted. For instance, the PhagoBurn study was established to evaluate the treatment of 51
Escherichia coli and Pseudomonas aeruginosa burn wound infections using bacteriophages. It 52
was also the first prospective multicentric, randomized, single blind and controlled phage therapy 53
clinical trial according to both Good Manufacturing (GMP) and Good Clinical Practices (GCP 54
(https://globalclinicaltrialdata.com/trial/GCT022014-000714-65). Additional clinical trials which 55
seek to identify phage cocktails to treat burn wound infections are underway 56
(https://globalclinicaltrialdata.com/trial/GCT0104323475). Other clinical trials have evaluated 57
bacteriophages as prebiotics, which are defined as indigestible dietary components that promote 58
specific beneficial bacterial species (https://globalclinicaltrialdata.com/trial/GCT0103269617). 59
In addition, viruses can also predict disease risk (14). A recent study characterizing the stool 60
viromes of children at increased risk for type 1 diabetes showed that enterovirus B (EV-B) was 61
one of the most prevalent viruses in the children’s stool samples. The study also found that 62
children with prolonged shedding of the same EV-B serotype had higher odds of developing islet 63
autoimmunity compared to children negative for EV-B (14). These findings demonstrate the 64
potential applications of viruses for both disease treatment and prediction. 65
66
A number of the discoveries highlighted above have been facilitated through the field of 67
viral metagenomics. Studies demonstrating the current and future potential for viral 68
metagenomics revolve around key focus areas: (i) understanding the natural history of viruses, 69
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
4
particularly virus presence/absence and potential integration/lysis mechanisms; (ii) identifying 70
viral associations with disease, including unexplained illnesses; (iii) identifying potential novel 71
viral relationships with health and disease; (iv) identifying viral associations as risk factors for 72
disease; and (v) using viral metagenomics as a surveillance tool for animal, community, and 73
global health. 74
75
Biases in viral metagenomics 76
Although there are a number of applications for viral metagenomics, biases can be 77
introduced at many points along the process. From sample collection and processing, to data 78
generation and analysis, many choices and processes can impact data quality, interpretation, and 79
comparison. A viral metagenomics pipeline usually includes sample collection, sample 80
processing, sequencing, and bioinformatic analyses, similar to a microbiome pipeline (Figure 2). 81
In the following sections, several of the potential biases introduced at the various steps are 82
discussed. 83
84
Sample collection and storage 85
Sample collection practices are often dependent on sample type. For instance, collection 86
methods for samples such as saliva and urine might differ from those of stool, due to the nature 87
of the sample matrix and the amount of material needed for downstream processing. Once a 88
sample has been appropriately collected, a major factor that can challenge the recovery of viral 89
nucleic acid and viral community information is storage temperature. Although this has not been 90
widely assessed in the context of viral metagenomics, microbiome studies have shown a 91
significant impact of storage temperature on bacterial diversity and membership (15). Studies 92
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
5
assessing individual virus titers can also provide a sense of the impact of storage temperature on 93
virus viability for downstream applications. For instance, a study assessing the viability of 94
bacteriophage MS2 in wastewater found that only 20% of the initial titer was inactivated at 4°C, 95
compared to a reduction of 57% when storing the sample at -80°C after approximately 8 days 96
(16). Tailed phages, like those infecting Staphylococcus aureus, can be stored at -80°C for long-97
term preservation as long as a stabilizer is added (17). This usually ensures viability by 98
protecting phage tails. Other viruses, like Hepatitis C, can be inactivated after 5 days at room 99
temperature, after 6 months at 4°C, and can show a decrease of 15.6% after 5 days at −20°C 100
(18). While each virus demonstrates differing inactivation rates depending on the storage 101
conditions used, storage at -80°C is typically considered to be the gold standard for long-term 102
preservation of samples intended for microbiome analyses, and warmer storage temperatures can 103
significantly influence results (15). 104
105
Certain studies, particularly those where samples are collected in remote areas or samples 106
associated with outbreaks and pandemics, may require preservation buffers if there is no 107
immediate access to a -80°C freezer. For example, RNAlater (Ambion Inc, Austin, TX, USA), 108
has shown to preserve tissue samples in a similar manner as stored snap-frozen samples (19). 109
More specific to viruses, OMNIgene.GUT (DNA Genotek, Ottawa, Canada) and its preservation 110
buffer are known to preserve dsDNA viruses, including CrAssphage and bacteriophage T5, in a 111
manner similar to storing at -80°C (20). Experiments have shown a significant decrease in 112
bacteriophage T5, but not in CrAssphage, at room temperature after 14 days without preservation 113
buffers or other stabilization agents (20). Similar experiments have shown the ability of 114
OMNIgene.GUT to preserve both bacteriophage T5 and CrAssphage under different conditions 115
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
6
(20). These results indicate that, although some viruses may be persistent across different sample 116
storage conditions, as in the case of CrAssphage, other viruses, like bacteriophage T5, may be 117
more sensitive to storage temperature, duration, and other storage conditions. It is possible that 118
these findings may also translate to other viruses, including RNA viruses (e.g. severe acute 119
respiratory syndrome coronavirus 2 (SARS-CoV2), hepatitis C, and influenza), indicating that it 120
is important to select the most suitable stabilization conditions depending on one’s study and its 121
goals. References materials, for instance, may provide insights into the expected outcomes. In 122
addition, future studies are needed to further assess the effect of storage conditions on viruses 123
associated with urine, blood and cerebrospinal fluid. Importantly, addition of a preservation 124
buffer to maintain virome profiles may not always be necessary when storing samples at -80°C. 125
This may be the case for stool samples, where the majority of the viruses are temperate 126
bacteriophages and have shown to be resistant to profile changes after long-term preservation at -127
80°C (21). This may also be explained by the stability in bacterial community composition after 128
long-term storage at -80°C (22). 129
130
Increasing numbers of freeze/thaw cycles are known to affect the detection and 131
distribution of members within microbial communities, particularly bacteria and viruses (23). 132
Freeze/thaw cycles are common when samples of interest are utilized multiple times without 133
prior aliquoting and homogenization. Freeze/thaw can also happen during freezer failures, or 134
insufficient maintenance of cold chain during the shipping of the samples to a laboratory facility. 135
This may result in, but is not limited to changes in viral titer and increased cellular debris, which 136
may in turn affect downstream purification steps including host cell nucleic acid removal (24). 137
Although certain DNA and RNA viruses, including Hepatitis B and C, HIV, and SARS-CoV, 138
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
7
have shown to be stable after multiple freeze/thaw cycles involving freezing at -70°C or -80°C, 139
followed by thawing in a water bath at 25°C for 1h (Hepatitis B and C) or room temperature 140
(HIV) (25–27), the sensitivity of bacteria to freeze/thaw cycles may have the potential to impact 141
the recovery, diversity, and composition of bacteriophages, particularly prophages (21), and 142
additional work is needed in this space. Certain viruses are directly affected by freeze/thaw 143
cycles, while others are relatively stable. For example, intact influenza virus is dramatically 144
impacted by a single freeze–thaw cycle, reducing the concentration of filaments by almost half 145
(28). In contrast, extracted RNA from Influenza H1N1, demonstrates stability for up to 56 days 146
at −80°C or −20°C, or up to 9 freeze/thaw cycles (29). Similar experiments should be performed 147
on one’s virus(es) of interest to assess viability and nucleic acid stability preceding molecular 148
analyses. Likewise, the effects of storage conditions and freeze/thaw cycles need to be assessed 149
in novel and emerging viruses. 150
151
Sample processing 152
Sample pre-processing for viral metagenomics is crucial. Viral metagenomic protocols 153
may include a filtration/concentration step, where samples are centrifuged to remove any debris. 154
Centrifugation may be followed by serial filtrations using 0.45 and/or 0.22µm membranes, which 155
aim to remove cells larger than the pore size selected. In addition, filtration may be followed by a 156
concentration step using protein columns, as well as a purification step using cesium chloride 157
gradient ultracentrifugation, for example (9). Other methods include dithiothreitol treatment 158
(DTT) prior filtration and nucleic acid extraction, as well as CsCl gradient ultracentrifugation 159
after filtration and DNAse treatment. Results have shown that, while the CsCl gradient 160
ultracentrifugation method outperforms the DTT methods in removing host DNA, DTT methods 161
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
8
discriminate less against specific phage species and can yield more DNA (30). Application of 162
one or several of these practices can significantly influence results. For instance, a previous study 163
testing one or several treatments for virus isolation and purification in saliva samples showed 164
that recovery of the vaccinia virus is reduced after centrifugation, as well as when applying 165
filtration through 0.45µm membranes (31). One possible reason for the loss of the vaccinia WR 166
virus when using filtration is the large size of the virus, which has dimensions of approximately 167
360 × 270 × 250nm. Using 0.22µm and 0.45 µm membranes may have resulted in virus retention 168
due to the large size of the virus. Studies have suggested pre-treatments of the filters that include 169
passing an appropriate buffer, 10% fetal calf serum, veal infusion broth, or bovine albumin prior 170
filtering to decrease virus retention when using an appropriate sized filter (32). In the case of 171
bacteriophages, it is essential to remove any bacterial cells using centrifugation and filtration, 172
and to apply DNAse and RNAse treatments before nucleic acid extraction to remove 173
extracellular nucleic acids that may be host-associated. This is particularly important because 174
phage genes share a degree of homology to bacterial genes and, thus, may affect downstream 175
data interpretation. 176
177
The application of one or several of these steps may be useful, particularly when working 178
with samples that originate from sites with high-level of host contamination. Saliva, blood, 179
biopsies, skin swabs, and cerebrospinal fluid are several sample types known to have a high 180
degree of host burden. In such cases, host cell removal may aid in obtaining accurate profiles and 181
increase the viral signal. Alternatively, enrichment methods for specific viruses have shown 182
promising results. For instance, ViroCap was designed to enrich nucleic acids from 34 viral 183
families infecting vertebrate hosts (33). Other examples include VirCapSeq-VERT, targeting 184
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
9
over 200 viral taxa (34), hybrid-capture target enrichment using PCR-generated capture probes 185
(35), and, more recently, multiplex amplicon- and hybrid capture-based sequencing with ultra-186
high-throughput metatranscriptomics for SARS-CoV-2 studies (36). However, in cases of low-187
volume samples, as in the case of skin swabs, applying several of these steps may result in 188
sample loss. In such cases, deeper sequencing and post-sequencing host sequence removal may 189
provide actionable results with the caveat of lower evenness (e.g., Shannon diversity index), as in 190
the case of skin swab samples (2). In other cases, deeper sequencing may not be required due to 191
the low-diversity of the samples, but may be useful to increase genome coverage (37). Another 192
study evaluating 16 different concentration, extraction, and purification protocols showed that 193
tangential flow filtration (TFF), pyrophosphate in combination with sonication, and 194
ultracentrifugation in a sucrose gradient yielded significantly greater numbers of virus-like 195
particles (38). While there is no method or combination of methods known to provide the most 196
optimal results, sample pre-processing would need to be evaluated as needed for the particular 197
virus(es) of interest, although this may add additional biases. 198
199
Nucleic acid extraction and amplification of viral nucleic acids 200
Historically, viral nucleic acid extraction methods have been developed according to the 201
sample type and the virus(es) being targeted (39, 40). Commercially available methods now 202
facilitate viral nucleic acid extraction, and a number of them are intended for specific sample 203
types. For instance, a study evaluating the nucleic acid extraction efficiency from HeLa cells 204
spiked with four viruses, including the double-stranded DNA Epstein-Barr virus, double-205
stranded RNA Reovirus 3, single-stranded RNA Feline leukemia virus, and respiratory syncytial 206
virus, using eleven commercially available extraction kits based on silica membrane column, 207
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
10
magnetic beads, and precipitation-based extractions found that dual extraction methods may 208
provide improved sensitivity for recovering nucleic acids from viruses with specific biochemical 209
and biophysical characteristics (41). In another study testing three different nucleic acid 210
extraction methods in biofilms, including chloroform, tetrasodium pyrophosphate in combination 211
with sonication, and dithiothreitol (DTT), showed varying DNA yields that may be explained by 212
the mode of action of the chemicals used. For instance, chloroform may denature the lipid 213
envelopes surrounding viral capsids, as in the case of viruses from the Phycodnaviridae family, 214
including, for instance, MpoVs, which infect Micromonas polaris (38, 42). Results indicate that 215
nucleic acid extraction methods would need to be tested independently, ideally using reference 216
materials, which may help understand the biases that could be introduced at this stage. 217
218
Viral nucleic acid extraction may result in low to moderate yields, depending on the 219
method(s) used. For this reason, whole amplification of the nucleic acids may be considered. 220
One method, known as multiple displacement amplification (MDA) can result in significant 221
higher nucleic acid yields as it utilizes a high-fidelity enzyme (43). Other amplification methods 222
include sequence-independent single primer amplification (SISPA), which is a primer initiated 223
technique that requires target sequence modification preceding the logarithmic amplification of 224
the DNA (44). However, random amplification methods may result in the overrepresentation of 225
certain viruses including single-stranded (ss) DNA viruses. For instance, a study assessing the 226
effect of MDA found the overrepresentation of bacteriophage M13 in a mock community 227
composed of seven DNA viruses including bacteriophage lambda, vaccinia virus, phage phi29, 228
adenovirus, bacteriophage M13, mice minute virus p, and a porcine circovirus (31). This 229
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
11
suggests that, while amplification methods aid in obtaining higher nucleic acid yields, results 230
should be carefully interpreted, particularly in terms of abundances. 231
232
Sequencing technology 233
Although Illumina sequencing has dominated the field of metagenomics, other sequencing 234
platforms are also currently being used (e.g., Ion semiconductor sequencing platform). Yet, 235
combining outputs of various sequencing platforms may not be ideal when performing certain 236
meta-analyses (i.e. analyses combining datasets from different studies). In the case of viromes, 237
however, utilizing Illumina vs. Ion Torrent sequencing may not affect the diversity output. This 238
has been noted for cerebrospinal fluid, where viral alpha- and beta-diversity were not 239
significantly altered by the sequencing platform (6). Additional studies are needed to understand 240
the effect of the sequencing platform on more complex samples (e.g. stool and soil samples). 241
High-throughput sequencing is also increasingly being applied to clinical specimens aiming to 242
identify specific pathogenic viruses. In a study assessing the effect of the sequencing platform 243
and the sequencing kit using clinical specimens positive for various enteroviruses and 244
polioviruses showed that the number of viral reads will depend on the type of virus, as well as 245
the sequencing platform and sequencing kit. For instance, the number of enterovirus reads did 246
not seem to be affected by the sequencing platform or sequencing kit, but the number of 247
poliovirus reads was affected (45). The virus coverage was also affected by the sequencing 248
platform and sequencing kit, where the genome coverage was highest when using the Illumina 249
MiSeq 500 v2 kit compared to the Ion Torrent PGM kits (45). This seemed to be dependent on 250
the higher number of reads generated by the Illumina MiSeq 500 v2 kit compared to the Ion 251
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
12
Torrent PGM kits. Ideally, further studies assessing the effect of the sequencing platform and 252
sequencing kit may include viral reference materials. 253
254
Bioinformatic analyses - Host removal and assembly 255
Bioinformatic analyses in viral metagenomics may include several steps preceding 256
annotation. As with many microbiome studies, viral metagenomics research is accompanied by 257
the in-silico removal of host sequence (i.e. human, animal, or plant) post-sequencing. This helps 258
to ensure that the analysis consists primarily of viral sequences, which in turn may significantly 259
reduce the amount of computer power and time to process results. Following host sequence 260
removal, viral metagenomic sequence analysis approaches may include viral sequence assembly, 261
and, depending on the annotation tool used, differing results can be acquired as a function of the 262
assembler used and the degree of success achieved in the assembly process. This makes 263
metagenomic assembly particularly challenging for virome data, as it may result in fragmented 264
assemblies and, consequently, poor annotations. A previous study tested the ability of 16 265
different tools for sequence assembly in several sample types, including a viral mock community 266
comprised of 12 viral genomes, 10 of which were at equal abundance (9.82% relative 267
abundance/virus) and 2 were ssDNA genomes (0.92% relative abundance/virus) (46) (Table 1). 268
For this mock community, specifically, particular assemblers, including CLC, Geneious, SPAdes 269
and VICUNA, were able to detect all 12 genomes. However, a number of false positives (i.e., 270
alignment to a number of reference genomes) were also identified. Other assemblers, 271
particularly, Velvet and MetaVelvet, generated no false positives, but failed to assemble three 272
genomes. In contrast, ABySS generated a large number of false positives and failed to assemble 273
four to six genomes, depending on the k-mer setting used. The assemblers IDBA UD and Ray 274
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
13
Meta outperformed the other assemblers with an equal number of contigs to genomes, followed 275
by MEGAHIT and SPAdes (46). This illustrates the biases introduced when assembling viral 276
sequences and the ability of reference materials to determine the performance of assembly tools. 277
Reference materials may be used in parallel to the sample of interest to determine the effect of 278
assemblers in detecting the expected viruses, as well as the specificity, including the number of 279
false positives and the number of false negatives. This may also need to be applied as novel 280
assemblers become available. 281
282
Bioinformatic analyses - Annotation tools 283
Annotation tools can also impact the outcomes of a viral metagenomics study. A number 284
of annotation tools specific for viruses are available, which can be alignment- or k-mer-based 285
(47). Since a number of annotation tools will continue to be available, it is important to evaluate 286
the most suitable annotation tool based on specific research needs. For instance, a study 287
assessing the viral content of benthic deep-sea samples using BLAST, MG-RAST, NBC, 288
VMGAP, MetaVir, and VIROME showed that the BLAST tools, followed by MetaVir and 289
VMGAP, provided the most reliable results. In addition, while tBLASTx, MetaVir, VMGAP and 290
VIROME showed a similar efficiency of sequence annotation, MetaVir and tBLASTx identified 291
a higher number of viral strains (48). Another tool, known as VirMAP employs multiple methods 292
that include a combination of de-novo assembly and mapping-based strategies to taxonomically 293
classify sequences (49). While these tools possess a number of advantages, most are database-294
reliant and not necessarily suitable for novel virus discovery. For this reason, another more 295
recent tool known as VIBRANT (Virus Identification By iteRative ANnoTation) enables both 296
virus identification and discovery utilizing machine learning and protein similarity approaches. 297
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
14
Importantly, VIBRANT highlights viral auxiliary metabolic genes and metabolic pathways, 298
which is usually not covered by other annotation tools (50). Another tool known as 299
DeepVirFinder is a reference and alignment-independent machine learning method for 300
identifying viral sequences in metagenomics using deep learning (51). Performance of current 301
and future viral taxonomic and functional annotation tools can be assessed using reference 302
materials. 303
304
Assessing biases in viral metagenomics 305
Although several approaches exist for assessing the biases introduced at one or more of 306
the steps described above, references materials represent a standardized manner to benchmark 307
pipelines, reagents, and bioinformatic analyses. Reference materials also represent an approach 308
to assess biases that could be introduced even after standardization (i.e. technician variation, 309
shipping issues, power shortages, etc.). However, the practice of including reference materials 310
arose relatively recently as an approach to address the reliability of Illumina sequencing as a 311
substitute of 454 sequencing (52). Specifically, the term “mock community” arose to refer to a 312
mix of bacterial DNA (52), and is currently applied to refer to a mix of bacterial or fungal cells 313
or DNA, as well as viral particles or nucleic acids in specific concentrations. A number of these 314
reference materials include mock communities composed of organisms with diverse structural 315
(e.g., cell-wall and morphology) and genomic characteristics (e.g., GC content and genome size), 316
mock communities mimicking various body sites, organisms intended to be spiked into a sample, 317
as well as homogenized stool samples. 318
319
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
15
Although the importance of reference materials is increasingly being acknowledged, 320
there is room for improvement in the field. For instance, only 30% of articles published in 2018 321
in two broadly-read and well-cited microbiome and microbiology journals reported a negative 322
control, and only 10% reported results from positive controls (53). This suggests that there is the 323
need to implement and report the results of both negative and positive controls in microbiome 324
studies. Although such controls may include a variety of materials, reference materials, 325
specifically, can be used as positive controls and negative controls; yet, reference materials, 326
including mock communities, are not implemented on a regular basis in microbiome studies even 327
though they are available through research and academic institutions, as well as commercial 328
facilities. Most reference materials available are intended for microbiome analysis targeting the 329
bacterial fraction, and these have shown to be helpful in benchmarking nucleic acid extraction 330
methods (54), sequencing platforms and sequencing kits (55), assemblers, as well taxonomic and 331
functional classification tools (56, 57). A limited number of reference materials have been 332
developed for viral metagenomics research, and it is anticipated that additional materials may be 333
developed as viral metagenomic applications experience broader adoption. 334
335
Viral reference materials, as with bacterial and fungal reference materials, aid in 336
assessing biases introduced into a virome pipeline, as described above. Commercially available 337
reference materials, particularly viral mock communities from the American Type Culture 338
Collection (ATCC, Manassas, VA, USA) can be used to identify and potentially quantify biases 339
introduced at particular stages of a viral metagenomics pipeline, as highlighted in the previous 340
sections. ATCC’s viral mock communities are composed of equal concentrations of either 341
nucleic acids from five different viruses, or whole viruses. The mock communities possess 342
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
16
double-stranded (ds) DNA viruses, including enveloped viruses (herpesvirus), and unenveloped 343
viruses (adenovirus). The mock communities also possess RNA viruses, including a positive 344
sense ssRNA virus (Zika virus), negative sense ssRNA virus (Influenza B virus and Human 345
respiratory syncytial virus), and a dsRNA virus (Reovirus 3) (58). 346
347
Potential future applications of viral metagenomics and areas for growth 348
Although viral metagenomics has been widely used for virome characterization in 349
various sample types, the approach can be leveraged in a number of different ways to address 350
current and future virus discovery and tracking. Some of these include, but are not limited to: (i) 351
expansion of viral databases through virus discovery efforts; (ii) surveillance activities, 352
particularly among wildlife reservoirs or in the context of SARS-COV-2 in wastewater; (iii) 353
identification of emerging and re-emerging pathogens in veterinary medicine; (iv) identification 354
of novel viral relationships in health and disease, including unexplained illnesses; (v) 355
understanding the potential effects of novel treatments (e.g. fecal transplantation and phage 356
therapy) on the virome and microbiome; (vi) layering virome information onto metagenomic 357
studies to provide new biological and clinical insights; (vii) continued improvements related to 358
sample collection, stabilization, extraction and detection assays; (viii) testing new reagents and 359
approaches; and (ix) development of reference materials for viral metagenomics and clinical 360
applications (Figure 3). 361
362
The numerous potential (future and ongoing) applications of viral metagenomics are 363
increasingly being realized. In the areas of discovery, for instance, viral metagenomics has 364
shown increasing value. In a study evaluating the use of viral metagenomics for the identification 365
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
17
of viruses in bats showed that, while only <1.0% of the reads are viral, 83% are bacteriophage, 366
insect-borne viruses, and plant viruses, and 14% represent mammalian viruses (59). In addition, 367
while genome segments within a number of the mammalian viruses shared a high degree of 368
homology to known viruses, a number of these segments differed, suggesting that bats possess a 369
number of viruses that remain to be characterized (59). This approach may also translate to other 370
animals, particularly those harboring potential zoonotic viruses. More recent evidence has also 371
demonstrated the value of viral metagenomics in tracing potential recombination hosts for 372
SARS-CoV-2. Although it has been demonstrated that bats may have been the original reservoirs 373
of the novel coronavirus, genomic evidence has also shown that the pangolin may have served as 374
an intermediate host (60). This was also demonstrated using viral metagenomics, where a 375
coronavirus strain, similar to the human strain was identified from metagenomic datasets (61). 376
377
In addition to the application of viral metagenomics for the discovery and identification 378
of viruses, including SARS-CoV-2, the potential for surveillance activities remains largely 379
unexplored. Viral surveillance activities in various sample types rely largely on PCR- or panel-380
based techniques, which exhibit a number of advantages, including specificity and speed. Yet, 381
one drawback is that surveillance activities based on PCR- or panel-based techniques may be 382
limited to the virus(es) of interest. Viral metagenomics, on the other hand, exhibits the additional 383
advantage of the identification of a number of different viruses simultaneously, including novel 384
viruses. This suggests that viral metagenomics may be added to the toolbox of current methods 385
for virus surveillance in clinical and environmental samples (62). Similarly, viral metagenomics 386
has an increasing potential in virus surveillance in novel treatment methods. For instance, fecal 387
transplantation has shown to be effective in treating Clostridium difficile infections, but the 388
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
18
transmission of viruses from donor to recipient remains largely unexplored. In a study assessing 389
the viral contents after fecal transplantation from a donor to three pediatric recipients found that 390
most of the viruses that were transmitted were bacteriophages (63), which is in agreement with a 391
previous study assessing the viral contents of chemostat systems (64). This approach may also 392
apply to the identification and surveillance of other potentially pathogenic viruses in donor and 393
recipient stool samples. Surveillance activities using sequencing approaches may also be applied 394
in under-characterized environments, such as the built environment (e.g. hospitals and schools). 395
Most studies characterizing the built environment focus on the bacterial fraction (65). Thus, viral 396
metagenomics studies of the built environment in developed and developing countries, and in 397
both rural and urbanized regions are still needed. Results may provide information regarding risk 398
of infection of viral pathogens, as well as transmission dynamics and routes of transmission (e.g. 399
airborne, fomite, and water routes) in various built environments (65). In addition, a number of 400
these built environments possess a high prevalence of multidrug-resistant bacteria, which 401
represent a risk to public health. Bacteriophages have, therefore, been suggested as a means to 402
control multidrug-resistant bacteria in built environments (65). In this case, sequencing 403
approaches may aid in identifying any prophages or prophage remnants in the multidrug-resistant 404
bacterial genomes, which often result in superinfection resistance. 405
406
Another area where viral metagenomics may have potential future applications and the 407
potential to further develop is in identifying novel and broad viral relationships in health and 408
disease, including potentially unexplained illnesses. For instance, a study applying high-409
throughput sequencing in low-biomass samples, including cerebrospinal fluid, blood and throat 410
swabs found that viruses were identified in 32% of the samples (66). In parallel, conventional 411
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
19
virus diagnostic tests were performed and in multiple cases, the identified viruses were not 412
included in the selected routine diagnostic tests (66). Interestingly, application of viral 413
metagenomics resulted in the adjustment of a subject’s treatment after exclusion of a viral 414
infection (66). 415
416
Another potential application of viral metagenomics is the continued development and 417
refinement of viral reference materials. Development of novel viral reference materials may aid 418
to determine biases that can be introduced when testing current and novel reagents for viral 419
nucleic acid extraction, or conversion or RNA to cDNA, for instance. In addition, development 420
of novel viral reference materials can be used to validate current and novel surveillance assays 421
(e.g., qPCR assays for the detection of SARS-CoV2 in sewage and environmental samples). 422
Novel viral reference materials can also be used to validate a metagenomic assay, starting from 423
sample collection down to data analysis. In many cases when viral reference materials represent 424
the original virus strain, it could represent an opportunity to track virus evolution. Moving 425
forward, we anticipate that a number of viral reference materials will continue to be added to the 426
collection of those currently available. 427
428
References 429
1. Mukhopadhya I, Segal JP, Carding SR, Hart AL, Hold GL. 2019. The gut virome: the 430
‘missing link’between gut bacteria and host immunity? Therap Adv Gastroenterol 431
12:1756284819836620. https://doi.org/10.1177/1756284819836620. 432
2. Hannigan GD, Meisel JS, Tyldsley AS, Zheng Q, Hodkinson BP, SanMiguel AJ, Minot S, 433
Bushman FD, Grice EA. 2015. The Human Skin Double-Stranded DNA Virome: 434
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
20
Topographical and Temporal Diversity, Genetic Enrichment, and Dynamic Associations 435
with the Host Microbiome. MBio 6:e01578-15. DOI:10.1128/mBio.01578-15. 436
3. Abeles SR, Robles-Sikisaka R, Ly M, Lum AG, Salzman J, Boehm TK, Pride DT. 2014. 437
Human oral viruses are personal, persistent and gender-consistent. ISME J 8:1753–1767. 438
4. Santiago-Rodriguez TM, Ly M, Bonilla N, Pride DT. 2015. The human urine virome in 439
association with urinary tract infections. Front Microbiol 6:14. 440
https://doi.org/10.3389/fmicb.2015.00014. 441
5. Moustafa A, Xie C, Kirkness E, Biggs W, Wong E, Turpaz Y, Bloom K, Delwart E, 442
Nelson KE, Venter JC, Telenti A. 2017. The blood DNA virome in 8,000 humans. PLoS 443
Pathog 13: e1006292. https://doi.org/10.1371/journal.ppat.1006292. 444
6. Ghose C, Ly M, Schwanemann LK, Shin JH, Atab K, Barr JJ, Little M, Schooley RT, 445
Chopyk J, Pride DT. 2019. The Virome of Cerebrospinal Fluid: Viruses Where We Once 446
Thought There Were None. Front Microbiol 10:2061. 447
https://doi.org/10.3389/fmicb.2019.02061 448
7. Maqsood R, Rodgers R, Rodriguez C, Handley SA, Ndao IM, Tarr PI, Warner BB, Lim 449
ES, Holtz LR. 2019. Discordant transmission of bacteria and viruses from mothers to 450
babies at birth. Microbiome 7:156. https://doi.org/10.1186/s40168-019-0766-7. 451
8. Schulfer A, Santiago-Rodriguez TM, Ly M, Borin JM, Chopyk J, Blaser MJ, Pride DT. 452
2020. Fecal Viral Community Responses to High-Fat Diet in Mice. mSphere 5:1. 453
DOI: 10.1128/mSphere.00833-19 454
9. Ly M, Jones MB, Abeles SR, Santiago-Rodriguez TM, Gao J, Chan IC, Ghose C, Pride 455
DT. 2016. Transmission of viruses via our microbiomes. Microbiome 4:1. DOI 456
10.1186/s40168-016-0212-z 457
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
21
10. Ly M, Abeles SR, Boehm TK, Robles-Sikisaka R, Naidu M, Santiago-Rodriguez T, Pride 458
DT. 2014. Altered Oral Viral Ecology in Association with Periodontal Disease. MBio 5:3. 459
DOI: 10.1128/mBio.01133-14. 460
11. Lopetuso LR, Ianiro G, Scaldaferri F, Cammarota G, Gasbarrini A. 2016. Gut Virome and 461
Inflammatory Bowel Disease. Inflamm Bowel Dis 22:1708-1712. 462
https://doi.org/10.1097/MIB.0000000000000807. 463
12. Abeles SR, Ly M, Santiago-Rodriguez TM, Pride DT. 2015. Effects of long term 464
antibiotic therapy on human oral and fecal viromes. PLoS One 10: e0134941. 465
doi:10.1371/journal.pone.0134941. 466
13. Duan Y, Llorente C, Lang S, Brandl K, Chu H, Jiang L, White RC, Clarke TH, Nguyen K, 467
Torralba M, Shao Y, Liu J, Hernandez-Morales A, Lessor L, Rahman IR, Miyamoto Y, 468
Ly M, Gao B, Sun W, Kiesel R, Hutmacher F, Lee S, Ventura-Cots M, Bosques-Padilla F, 469
Verna EC, Abraldes JG, Brown RS, Vargas V, Altamirano J, Caballería J, Shawcross DL, 470
Ho SB, Louvet A, Lucey MR, Mathurin P, Garcia-Tsao G, Bataller R, Tu XM, Eckmann 471
L, van der Donk WA, Young R, Lawley TD, Stärkel P, Pride D, Fouts DE, Schnabl B. 472
2019. Bacteriophage targeting of gut bacterium attenuates alcoholic liver disease. Nature 473
575:505-511. https://doi.org/10.1038/s41586-019-1742-x. 474
14. Vehik K, Lynch KF, Wong MC, Tian X, Ross MC, Gibbs RA, Ajami NJ, Petrosino JF, 475
Rewers M, Toppari J, Ziegler AG, She JX, Lernmark A, Akolkar B, Hagopian WA, 476
Schatz DA, Krischer JP, Hyöty H, Lloyd RE. 2019. Prospective virome analyses in young 477
children at increased genetic risk for type 1 diabetes. Nat Med 25: 1865-1872. 478
https://doi.org/10.1038/s41591-019-0667-0. 479
15. van Zyl KN, Whitelaw AC, Newton-Foot M. 2020. The effect of storage conditions on 480
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
22
microbial communities in stool. PLoS One 15: e0227486. 481
https://doi.org/10.1371/journal.pone.0227486 482
16. Olson MR, Axler RP, Hicks RE. 2004. Effects of freezing and storage temperature on 483
MS2 viability. J Virol Methods 122: 147-152. 484
https://doi.org/10.1016/j.jviromet.2004.08.010. 485
17. Gonzalez-Menendez E, Fernandez L, Gutierrez D, Rodríguez A, Martínez B, GarcíaI P. 486
2018. Comparative analysis of different preservation techniques for the storage of 487
Staphylococcus phages aimed for the industrial development of phage-based antimicrobial 488
products. PLoS One 13: e0205728. https://doi.org/10.1371/journal.pone.0205728. 489
18. Halfon P, Khiri H, Gerolami V, Bourliere M, Feryn JM, Reynier P, Gauthier A, Cartouzou 490
G. 1996. Impact of various handling and storage conditions on quantitative detection of 491
hepatitis C virus RNA. J Hepatol 25: 307-311. https://doi.org/10.1016/S0168-492
8278(96)80116-4. 493
19. Kohl C, Wegener M, Nitsche A, Kurth A. 2017. Use of RNALater® preservation for 494
virome sequencing in outbreak settings. Front Microbiol 8:1888. 495
https://doi.org/10.3389/fmicb.2017.01888. 496
20. Le François B, Doukhanine E. 2019. OMNIgene®• GUT accurately captures and 497
stabilizes the human fecal dsDNA virome. White paper. 498
21. Shkoporov AN, Ryan FJ, Draper LA, Forde A, Stockdale SR, Daly KM, McDonnell SA, 499
Nolan JA, Sutton TDS, Dalmasso M, McCann A, Ross RP, Hill C. 2018. Reproducible 500
protocols for metagenomic analysis of human faecal phageomes. Microbiome 6:68. 501
https://doi.org/10.1186/s40168-018-0446-z. 502
22. Dorsaz S, Charretier Y, Girard M, Gaïa N, Leo S, Schrenzel J, Harbarth S, Huttner B, 503
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
23
Lazarevic V. 2020. Changes in Microbiota Profiles After Prolonged Frozen Storage of 504
Stool Suspensions. Front Cell Infect Microbiol 10:77. 505
https://doi.org/10.3389/fcimb.2020.00077. 506
23. Song SJ, Amir A, Metcalf JL, Amato KR, Xu ZZ, Humphrey G, Knight R. 2016. 507
Preservation Methods Differ in Fecal Microbiome Stability, Affecting Suitability for Field 508
Studies. mSystems 1:3. DOI: 10.1128/mSystems.00021-16. 509
24. Khan AS, Ng SHS, Vandeputte O, Aljanahi A, Deyati A, Cassart J-P, Charlebois RL, 510
Taliaferro LP. 2017. A Multicenter Study To Evaluate the Performance of High-511
Throughput Sequencing for Virus Detection. mSphere 2:5. DOI: 10.1128/mSphere.00307-512
17. 513
25. Krajden M, Minor JM, Rifkin O, Comanor L. 1999. Effect of multiple freeze-thaw cycles 514
on hepatitis B virus DNA and hepatitis C virus RNA quantification as measured with 515
branched-DNA technology. J Clin Microbiol 37: 1683-1686. 516
DOI: 10.1128/JCM.37.6.1683-1686.1999 517
26. Griffith BP, Rigsby MO, Garner RB, Gordon MM, Chacko TM. 1997. Comparison of the 518
Amplicor HIV-1 Monitor test and the nucleic acid sequence-based amplification assay for 519
quantitation of human immunodeficiency virus RNA in plasma, serum, and plasma 520
subjected to freeze-thaw cycles. J Clin Microbiol 35: 3288-3291. 521
27. Chan KH, Poon LLLM, Cheng VCC, Guan Y, Hung IFN, Kong J, Yam LYC, Seto WH, 522
Yuen KY, Peiris JSM. 2004. Detection of SARS Coronavirus in Patients with Suspected 523
SARS. Emerg Infect Dis 10: 294. DOI: 10.3201/eid1002.030610 524
28. Hirst JC, Hutchinson EC. 2019. Single-particle measurements of filamentous influenza 525
virions reveal damage induced by freezing. J Gen Virol 100:1631. DOI: 526
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
24
10.1099/jgv.0.001330 527
29. Granados A, Petrich A, McGeer A, Gubbay JB. 2017. Measuring influenza RNA quantity 528
after prolonged storage or multiple freeze/thaw cycles. J Virol Methods 247:45-50. 529
https://doi.org/10.1016/j.jviromet.2017.05.018 530
30. Kleiner M, Hooper L V., Duerkop BA. 2015. Evaluation of methods to purify virus-like 531
particles for metagenomic sequencing of intestinal viromes. BMC Genomics 16:7. DOI 532
10.1186/s12864-014-1207-4. 533
31. Parras-Moltó M, Rodríguez-Galet A, Suárez-Rodríguez P, López-Bueno A. 2018. 534
Evaluation of bias induced by viral enrichment and random amplification protocols in 535
metagenomic surveys of saliva DNA viruses. Microbiome 6:119. 536
https://doi.org/10.1186/s40168-018-0507-3. 537
32. Ver BA, Melnick JL, Wallis C. 1968. Efficient Filtration and Sizing of Viruses with 538
Membrane Filters. J Virol 2:21-25. 539
33. Wylie TN, Wylie KM, Herter BN, Storch GA. 2015. Enhanced virome sequencing using 540
targeted sequence capture. Genome Res 25:1910-1920. DOI: 10.1101/gr.191049.115 541
34. Briese T, Kapoor A, Mishra N, Jain K, Kumar A, Jabado OJ, Ian Lipkina W. 2015. 542
Virome capture sequencing enables sensitive viral diagnosis and comprehensive virome 543
analysis. MBio 6:5. DOI: 10.1128/mBio.01491-15. 544
35. Duncavage EJ, Magrini V, Becker N, Armstrong JR, Demeter RT, Wylie T, Abel HJ, 545
Pfeifer JD. 2011. Hybrid capture and next-generation sequencing identify viral integration 546
sites from formalin-fixed, paraffin-embedded tissue. J Mol Diagnostics 13:325-333. 547
https://doi.org/10.1016/j.jmoldx.2011.01.006. 548
36. Xiao, M., Liu, X., Ji, J., Li, M., Li, J., Yang, L., Sun, W., Ren, P., Yang, G., Zhao, J. 2020. 549
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
25
Multiple approaches for massively parallel sequencing of SARS-CoV-2 genomes directly 550
from clinical samples. Genome Med 12:57. 551
37. Illingworth CJR, Roy S, Beale MA, Tutill H, Williams R, Breuer J. 2017. On the effective 552
depth of viral sequence data. Virus Evol 3:2. https://doi.org/10.1093/ve/vex030. 553
38. Bekliz M, Brandani J, Bourquin M, Battin TJ, Peter H. 2019. Benchmarking protocols for 554
the metagenomic analysis of stream biofilm viromes. PeerJ 7: e8187. 555
https://doi.org/10.7717/peerj.8187. 556
39. Bergallo M, Costa C, Gribaudo G, Tarallo S, Baro S, Ponzi AN, Cavallo R. 2006. 557
Evaluation of six methods for extraction and purification of viral DNA from urine and 558
serum samples. New Microbiol 29:111-119. 559
40. Espy MJ, Patel R, Paya C V., Smith TF. 1995. Comparison of three methods for 560
extraction of viral nucleic acids from blood cultures. J Clin Microbiol 33:41-44. 561
41. Sathiamoorthy S, Malott RJ, Gisonni-Lex L, Ng SHS. 2018. Selection and evaluation of 562
an efficient method for the recovery of viral nucleic acid extraction from complex 563
biologicals. npj Vaccines 3:1-6. https://doi.org/10.1038/s41541-018-0067-3. 564
42. Van Etten JL, Dunigan DD, Nagasaki K, Schroeder DC, Grimsley N, Brussaard CPD, 565
Nissimov JI. 2020. Phycodnaviruses (Phycodnaviridae)☆. DOI:10.1016/B978-0-12-566
809633-8.21291-0. 567
43. Shoaib M, Baconnais S, Mechold U, Le Cam E, Lipinski M, Ogryzko V. 2008. Multiple 568
displacement amplification for complex mixtures of DNA fragments. BMC Genomics 569
9:1-14. https://doi.org/10.1186/1471-2164-9-415. 570
44. Reyes GR, Kim JP. 1991. Sequence-independent, single-primer amplification (SISPA) of 571
complex DNA populations. Mol Cell Probes 5:473-481. https://doi.org/10.1016/S0890-572
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
26
8508(05)80020-9. 573
45. Marine RL, Magaña LC, Castro CJ, Zhao K, Montmayeur AM, Schmidt A, Diez-Valcarce 574
M, Ng TFF, Vinjé J, Burns CC. 2020. Comparison of Illumina MiSeq and the Ion Torrent 575
PGM and S5 platforms for whole-genome sequencing of picornaviruses and caliciviruses. 576
J Virol Methods 113865. https://doi.org/10.1016/j.jviromet.2020.113865. 577
46. Sutton TDS, Clooney AG, Ryan FJ, Ross RP, Hill C. 2019. Choice of assembly software 578
has a critical impact on virome characterisation. Microbiome 7:12. 579
https://doi.org/10.1186/s40168-019-0626-5. 580
47. Santiago-Rodriguez TM, Hollister EB. 2019. Human Virome and Disease: High-581
Throughput Sequencing for Virus Discovery, Identification of Phage-Bacteria Dysbiosis 582
and Development of Therapeutic Approaches with Emphasis on the Human Gut. Viruses 583
11:656. https://doi.org/10.3390/v11070656. 584
48. Tangherlini M, Dell’Anno A, Zeigler Allen L, Riccioni G, Corinaldesi C. 2016. Assessing 585
viral taxonomic composition in benthic marine ecosystems: Reliability and efficiency of 586
different bioinformatic tools for viral metagenomic analyses. Sci Rep 6:28428. 587
https://doi.org/10.1038/srep28428. 588
49. Ajami NJ, Wong MC, Ross MC, Lloyd RE, Petrosino JF. 2018. Maximal viral 589
information recovery from sequence data using VirMAP. Nat Commun 9:1-9. 590
https://doi.org/10.1038/s41467-018-05658-8. 591
50. Anantharaman K, Kieft K, Zhou Z. 2019. VIBRANT: Automated recovery, annotation 592
and curation of microbial viruses, and evaluation of virome function from genomic 593
sequences. bioRxiv, 855387. DOI: https://doi.org/10.1101/855387. 594
51. Ren J, Song K, Deng C, Ahlgren NA, Fuhrman JA, Li Y, Xie X, Poplin R, Sun F. 2020. 595
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
27
Identifying viruses from metagenomic data using deep learning. Quant Biol 8:64-77. 596
https://doi.org/10.1007/s40484-019-0187-4. 597
52. Kozich JJ, Westcott SL, Baxter NT, Highlander SK, Schloss PD. 2013. Development of a 598
dual-index sequencing strategy and curation pipeline for analyzing amplicon sequence 599
data on the miseq illumina sequencing platform. Appl Environ Microbiol 79: 5112-5120. 600
DOI: 10.1128/AEM.01043-13. 601
53. Hornung BVH, Zwittink RD, Kuijper EJ. 2019. Issues and current standards of controls in 602
microbiome research. FEMS Microbiol Ecol 95: fiz045. 603
https://doi.org/10.1093/femsec/fiz045. 604
54. Leigh Greathouse K, Sinha R, Vogtmann E. 2019. DNA extraction for human microbiome 605
studies: The issue of standardization. Genome Biol 20:212. 606
https://doi.org/10.1186/s13059-019-1843-8. 607
55. Onywera H, Meiring TL. 2020. Comparative analyses of Ion Torrent V4 and Illumina V3-608
V4 16S rRNA gene metabarcoding methods for characterization of cervical microbiota: 609
taxonomic and functional profiling. Sci African: e00278. 610
https://doi.org/10.1016/j.sciaf.2020.e00278. 611
56. Ye SH, Siddle KJ, Park DJ, Sabeti PC. 2019. Benchmarking Metagenomics Tools for 612
Taxonomic Classification. Cell 178: 779-794. https://doi.org/10.1016/j.cell.2019.07.010. 613
57. Pollock J, Glendinning L, Wisedchanwet T, Watson M. 2018. The Madness of 614
Microbiome: Attempting To Find Consensus “Best Practice” for 16S Microbiome Studies. 615
Appl Environ Microbiol 84:7. DOI: 10.1128/AEM.02627-17. 616
58. Lopera, Juan; Brenton, Briana; Sohn, Jung-Woo; King, Stephen; Santiago-Rodriguez, 617
Tasha M; Hollister, Emily B; Wong, Matthew C; Ajami, Nadim; Wilder C. 2019. 618
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
28
Development and Evaluation of Next-Generation Sequencing Standards for Virome 619
Research. Manassas, VA. White paper. 620
59. Mishra N, Fagbo SF, Alagaili AN, Nitido A, Williams SH, Ng J, Lee B, Durosinlorun A, 621
Garcia JA, Jain K, Kapoor V, Epstein JH, Briese T, Memish ZA, Olival KJ, Ian Lipkin W. 622
2019. A viral metagenomic survey identifies known and novel mammalian viruses in bats 623
from Saudi Arabia. PLoS One 14: e0214227. 624
https://doi.org/10.1371/journal.pone.0214227. 625
60. Zhang T, Wu Q, Zhang Z. 2020. Probable Pangolin Origin of SARS-CoV-2 Associated 626
with the COVID-19 Outbreak. Curr Biol 30: 1346-1351.e2. 627
https://doi.org/10.1016/j.cub.2020.03.022. 628
61. Wong MC, Cregeen SJJ, Ajami NJ, Petrosino JF. 2020. Evidence of recombination in 629
coronaviruses implicating pangolin origins of nCoV-2019. bioRxiv. 630
https://doi.org/10.1101/2020.02.07.939207. 631
62. Kumar R, Nagpal S, Kaushik S, Mendiratta S. 2020. COVID-19 diagnostic approaches: 632
different roads to the same destination. VirusDisease 31, 97–105. 633
https://doi.org/10.1007/s13337-020-00599-7. 634
63. Chehoud C, Dryga A, Hwang Y, Nagy-Szakal D, Hollister EB, Luna RA, Versalovic J, 635
Kellermayer R, Bushman FD. 2016. Transfer of viral communities between human 636
individuals during fecal microbiota transplantation. MBio 7:2. DOI: 10.1128/mBio.00322-637
16. 638
64. Santiago-Rodriguez TM, Ly M, Daigneault MC, Brown IHL, McDonald JAK, Bonilla N, 639
Vercoe EA, Pride DT. 2015. Chemostat culture systems support diverse bacteriophage 640
communities from human feces. Microbiome 3:1-16. https://doi.org/10.1186/s40168-015-641
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
29
0124-3. 642
65. Prussin AJ, Belser JA, Bischoff W, Kelley ST, Lin K, Lindsley WG, Nshimyimana JP, 643
Schuit M, Wu Z, Bibby K, Marr LC. 2020. Viruses in the Built Environment (VIBE) 644
meeting report. Microbiome 8:1. https://doi.org/10.1186/s40168-019-0777-4. 645
66. Kufner V, Plate A, Schmutz S, Braun DL, Günthard HF, Capaul R, Zbinden A, Mueller 646
NJ, Trkola A, Huber M. 2019. Two years of viral metagenomics in a tertiary diagnostics 647
unit: Evaluation of the first 105 cases. Genes (Basel)10:661. 648
https://doi.org/10.3390/genes10090661. 649
650
Figure legends 651
Figure 1. Number of microbiome and virome publications. Panel A shows the number of 652
microbiome publications since 2000 (blue line) and virome publications since 2006 (green line). 653
Panel B shows an expanded graph of the number of viral metagenomic studies since 2006. 654
Figure 2. Description of a viral metagenomics pipeline. A viral metagenomics pipeline may 655
include sample collection, processing, sequencing and bioinformatics. Figure shows several of 656
the steps that may add biases to the results. 657
Figure 3. Potential applications of viral metagenomics in ongoing and future fields related to 658
database expansion, surveillance, identification of viral relationships with health and disease, 659
reagents development, among others. 660
661
Authors Bios: 662
Emily Hollister received her Ph.D. in Molecular and Environmental Plant Sciences from Texas 663
A&M University and completed her postdoctoral training in Molecular Microbial Ecology at 664
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
30
Texas A&M as well. She currently serves as the Director, R&D, Computational Biology for 665
Diversigen, where she and her team focus on the development and implementation of novel 666
solutions for 'omics data analysis. She has pursued a variety of research topics related to the 667
microbiome and human health over the course of the last 10 years. 668
669
Tasha M. Santiago-Rodriguez received her Ph.D. in Biology with emphasis on Public Health 670
Water Microbiology from the University of Puerto Rico characterizing bacteriophages as 671
markers of human fecal contamination. She completed her postdoctoral training in virome and 672
ancient microbiomes at the University of California, San Diego, and California Polytechnic State 673
University, San Luis Obispo, respectively. She is currently an R&D Bioinformatician at 674
Diversigen, and has been in the microbiome field for over 7 years. 675
676
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
Table 1. Assemblers performance. Performance was previously evaluated using, among many factors, the number of false positives,
false negatives, true positives, number of contigs and sensitivity. Links to the assembler sources are also shown. Modified from [46].
Assembler
False
positives
False
negatives
True
positive
Number of
contigs Sensitivity Source
ABySS (v2.0.2) (k-mer 63) 52 4 8 61 66.67 http://www.bcgsc.ca/downloads/abyss/
ABySS (v2.0.2) (k-mer 127) 50 6 6 56 50 http://www.bcgsc.ca/downloads/abyss/
CLC (v5.0.5) 1143 0 12 1299 100 https://www.qiagenbioinformatics.com/products/clc-assembly-cell/
Geneious (v11.0.3) 53 0 12 65 100 https://www.geneious.com/features/assembly-mapping/
IDBA UD (v1.1.1) 0 0 12 12 100 https://i.cs.hku.hk/~alse/hkubrg/projects/idba_ud
MEGAHIT (v1.1.1-2) 0 0 12 13 100 https://github.com/voutcn/megahit
MetaVelvet (v1.2.02) 0 3 9 26 75 https://metavelvet.dna.bio.keio.ac.jp/
MIRA (v4.0.2) 0 0 12 89 100 http://www.chevreux.org/mira_downloads.html
Ray Meta (v2.3.0) 0 0 12 12 100 http://denovoassembler.sourceforge.net/
SOAPdenovo2 (v2.04) 2 0 12 23 100 http://soap.genomics.org.cn/soapdenovo.html
SPAdes(v3.10.0) 0 0 12 14 100 http://cab.spbu.ru/software/spades/
SPAdes meta (v3.10.0) 0 0 12 14 100 http://cab.spbu.ru/software/spades/ (variation of SPAdes applied with flag)
SPAdes sc 1513 0 12 1527 100 http://cab.spbu.ru/software/spades/ (variation of SPAdes applied with flag)
SPAdes sc careful 0 0 12 15 100 http://cab.spbu.ru/software/spades/ (variation of SPAdes applied with flag)
Velvet (v1.2.10) 0 3 9 26 75 https://www.ebi.ac.uk/~zerbino/velvet/
VICUNA (v1.3) 4969 0 12 5385 100 https://github.com/broadinstitute/mvicuna
on May 21, 2021 by guest
http://aem.asm
.org/D
ownloaded from
